How police officers use algorithms to predict (and prevent) crimes

Law enforcement agencies around the country are beginning to use computers to predict when and where crimes are most likely to occur.
Written by Sarah Korones, Contributor

What if police officers had the power to predict both the time and the exact location of a crime before it happened? Such knowledge would allow them to be on the scene ahead of time, stopping illegal incidents before they ever had the chance to play out.

Law enforcement officials don't yet possess these superhero-like abilities themselves, but police in Los Angeles, Calif. can now predict when and where crimes such as car theft and burglary will occur thanks to a computer program called PredPol.

Developed by a startup based in Santa Cruz, Calif., PredPol pinpoints the locations at which crimes are most likely to occur by gathering data and applying algorithms. Initially developed to predict earthquakes, the software has already been tested in Santa Cruz and Los Angeles.

Now, the Los Angeles Police Department has announced that it will deploy the predictive policing software on a much larger scale, using it to monitor at least six different precincts. The rollout follows initial tests in the city's Foothill precinct, after which police officers reported a 25 percent drop in reported burglaries in the neighborhoods to which the program directed them.

MIT's Technology Review explains how the software works:

The inputs are straightforward: previous crime reports, which include the time and location of a crime. The software is informed by sociological studies of criminal behavior, which include the insight that burglars often ply the same area.

The system produces, for each patrol shift, printed maps speckled with red boxes, 500 feet on each side, suggesting where property crimes specifically, burglaries and car break-ins and thefts are statistically more likely to happen. Patterns detected over a period of several years as well as recent clusters figure in the algorithm, and the boxes are recalibrated for each patrol shift based on the timeliest data.

Once given a map of hotspots, officers are encouraged to spend time in the marked areas. The emphasis is less on actually solving crimes and arresting people, and more on preventing crimes by establishing a high profile police presence in high-risk areas.

By letting the computers take care of the detective work, officers are able to spend more time on the streets an appealing asset for cash-strapped agencies who are already stretching their personnel thin and are looking for computers to handle some of the work.

Not everyone is thrilled with the idea of stopping crimes before they happen, however. Predictive policing, some say, could lead to racial profiling or reinforcing stereotypes. As Smithsonian.com notes, once a computer identifies a risky area, the bar for what is considered suspicious behavior may lower significantly.

[via Smithsonian via Technology Review]

Image: PredPol, Jametiks/Flickr

This post was originally published on Smartplanet.com

Editorial standards